Minutes, 5 Dec 2002 WS Description telcon [html]

WS Description WG Teleconference
5 Dec 2002


See also: IRC log <http://www.w3.org/2002/12/05-ws-desc-irc>  (members only)


Attendees


Present:
 David Booth            W3C
 Allen Brookes          Rogue Wave Software
 Roberto Chinnici       Sun Microsystems
 Glen Daniels           Macromedia
 Youenn Fablet          Canon
 Dietmar Gaertner       Software AG
 Martin Gudgin          Microsoft
 Tom Jordahl            Macromedia
 Jacek Kopecky          Systinet
 Philippe Le Hégaret    W3C
 Steve Lind             AT&T
 Kevin Canyang Liu      SAP
 Lily Liu               webMethods
 Jonathan Marsh         Chair (Microsoft)
 Dale Moberg            Cyclone Commerce
 Jean-Jacques Moreau    Canon
 Arthur Ryman           IBM
 Adi Sakala             IONA Technologies
 Jeffrey Schlimmer      Microsoft
 Igor Sedukhin          Computer Associates
 Jerry Thrasher         Lexmark
 William Vambenepe      Hewlett-Packard
 Don Wright             Lexmark
 Joyce Yang             Oracle
 Prasad Yendluri        webMethods, Inc.
 Barbara Zengler        DaimlerChrysler Research and Technology

Regrets:
 Steve Graham           Global Grid Forum
 Sandeep Kumar          Cisco Systems
 Amy Lewis           Tibco
 Don Mullen           Tibco
 Waqar Sadiq            Electronic Data Systems
 Sanjiva Weerawarana    IBM

Chair: JMarsh 

Scribe: dbooth


Contents


*	Agenda <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#agenda#agenda>  Items 

1.	Approval <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item01#item01>  of Minutes 
2.	Review <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item02#item02>  of Action Items 
3.	Administrivia <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item03#item03>  
4.	Web <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item04#item04>  Service definition (Sanjiva) 
5.	Publication <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item05#item05>  issues 
6.	Properties <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item06#item06>  and features 
7.	Proposal: <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item07#item07>  MEP support in operations 
8.	Removing <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item08#item08>  message. Roberto's proposal 
9.	Output <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#item09#item09>  operations 

*	Summary <file:///C:\Documents%20and%20Settings\jmarsh\Local%20Settings\Temporary%20Internet%20Files\OLKA5\minutes-irc-021205.htm#newActions#newActions>  of Action Items 

  _____  


Approval of Minutes


Scribe: Minutes approved: http://lists.w3.org/Archives/Public/www-ws-desc/2002Nov/0095.html 


Review of Action Items


Scribe: ACTION: 2002-11-11: Arthur will submit targetNamespace/simplifiedNun text to editors to be included in part 1 as non-normative appendix, including a note regarding non backward compatibility with wsdl 1.1. -- DONE 

<Marsh> http://lists.w3.org/Archives/Public/www-ws-desc/2002Dec/0021.html 

Scribe: ACTION: 2002-11-12: Paco will write two options for naming faults: schema vs WSDL. -- PENDING 
... ACTION: 2002-11-12: Roberto will try and come up with another proposal for eliminating message, the discussion goes to email or the next f2f. -- PENDING 
... ACTION: 2002-11-12: Marsh to contact the XMLP WG (and later the wider public) speaking about the issues around removing the use attribute. -- PENDING. 

JMarsh: Sent msg to DavidFallside. Haven't heard back yet. 

Scribe: ACTION: 2002-11-12: Glen and Paco to chase the Global Grid Forum WRT services implementing a single portType. -- PENDING 
... ACTION: 2002-11-13: Glen to write the 2 proposals for specifying Binding properties and send them to the list (1 = add <*:property> to <binding>, 2 = move protocol binding stuff including properties to ports.) -- REVISED 
... ACTION: Glen to write up a description of the issues surrounding property description in WSDL. (This action replaces Glen's previous action item.) 
... 2002-11-21: Don Mullen to detail changes/addition necessary to unify SOAP and WSDL MEPs. -- PENDING 
... 2002-11-21: Jonathan to refer R120 text to TAG, referencing TAG issue fragmentinXML-28, when that text appears in the draft. -- PENDING 


Administrivia


JMarsh: Welcome new member: Lily Liu of webMethods. 

<Philippe> http://www.w3.org/2002/12/WSf2fJanLogistics.html 

Philippe: On F2F logistics, Don is proposing some hotels. He is proposing to book a block of rooms in the Marriott, but it's more expensive than others: $180 to 220. 
... Marriott has high speed internet; the others don't. 

<Philippe> Marriott is at $180 to 220 
... others are at $130-140 

Scribe: AGREED: Marriott sounds fine. 

JMarsh: Wednesday we could meet concurrently with the Arch group, or meet for a half day. 

dbooth: Half day sounds like a good idea to me. 

Scribe: AGREED: JMarsh will suggest to the Arch WG that the WS Desc WG ends at 12 on Wednesday. 

JMarsh: Glen requested that action items be posted to a group page. 

dbooth: All pending action items should always be listed in the minutes, so perhaps it is enough to look there. 
... I suggest that whoever sends out the minutes should specifically copy anyone who has an action item. 

Scribe: ACTION: dbooth to put URL for current minutes on group page [Done: http://www.w3.org/2002/ws/desc/#agenda ] 

dbooth: (Group congratulates Sanjiva on his new family member!) 


Web Service definition (Sanjiva)


JMarsh: Sanjiva seems to have raised a new issue. 

<Marsh> http://lists.w3.org/Archives/Public/www-ws-desc/2002Dec/0000.html 

JMarsh: We'll leave it to the editors to address, since Sanjiva is not on the call. 
... Also a question of types and arrays. Is it an issue yet? 

Jacek: Not an issue yet. 


Publication issues


JMarsh: New drafts expected for review by Dec 6th. 
... Is that still realistic? 

Gudge: No. 
... I think Tuesday 2002-12-10 is doable. 

JMarsh: We'll still shoot for publication before the end of the year. 
... Regarding capitalization, we need to adopt a consistent practice. 
... Options: web service, Web service, Web Service. 

Joyce: Lower case w is sometimes used for description. 

Scribe: Straw poll on capitalization 

<Marsh> Option 1: web service 
... Option 2: Web service 
... Option 3: Web Service 

<youenn> +1 for Option 1 

Scribe: Straw poll results: 

Joyce: Option 1: 3 
... Option 2: 2 
... Option 3: 14 
... Abstain: 2 

Scribe: AGREED: Incorporate "Web Service" in our draft, and let the WS Arch group know the straw poll results and let them tell us differently if they want. 


Properties and features


JMarsh: Glen still has an action item to raise some issues in prep for our F2F. 

Glen: What we're trying to do is provide for the max usability in WSDL for things that will arise from Soap extensibility model (and perhaps more if Arch WG adopts it). 
... But there aren't many use cases for it. 
... We dont' have a one-way MEP defined. In XMLP we discussed the concept of having both request-response over HTTP, and using Soap headers to do it over other things like UDP. 
... But we didn't actually get there and demonstrate the whole thing. 
... I don't want to restrict the ways people will use WSDL. 
... Where in WSDL do you want to talk about these things? 
... Specifying the use is pretty clear cut. 
... But there are a number of other issues. 
... For example, putting the feature inside a particular module. 
... The properties are an abstract bag that all can access. 
... So you might have 4 Soap modules that all have a UserID property. 
... But another module on non-repudiation might refer to exactly the same thing, so putting them inside a feature/module might not make sense. 
... The simple way is to say "We're using this binding with this URI and this module with this URI". 
... But once we talk about how the properties are constrained, we have to think harder about where to put them. 

Scribe: (Glen will write it up) 

Glen: We also need to finish the merge on the current proposals. 
... I think we can do that on the high level, with open issues on some property constraints. 

Scribe: Glen's slice: http://lists.w3.org/Archives/Public/www-ws-desc/2002Sep/0004.html 
... Amy's slice: http://lists.w3.org/Archives/Public/www-ws-desc/2002Nov/0083.html 

Glen: The feature/binding stuff that Amy proposes maybe should be considered separately. 


Proposal: MEP support in operations 


Glen: It would be nice to see some fully fleshed-out scenarios. 

Scribe: http://lists.w3.org/Archives/Public/www-ws-desc/2002Nov/0085.html 


Removing message. Roberto's proposal 


Scribe: Roberto's proposal: http://lists.w3.org/Archives/Public/www-ws-desc/2002Nov/0035.html 

Roberto: I hope to have a revision by the next call. 


Output operations


Scribe: Amy's Day Trader application: http://lists.w3.org/Archives/Public/www-ws-desc/2002Nov/0052.html 

Glen: We're all over the place on this. 

TomJ: Is it okay to retain the status quo (keep output operations)? 

<jeffsch> Yes 
... +1 to additional clarification as needed 

Prasad: If we keep them we should at least define them completely. 

JMarsh: Three options: 1. Leave them incompletely described, 2. Describe them completely, 3. Remove them. 

Joyce: I dont' think removing them is an option, because people are using them. 

TomJ: But others say they aren't well specified. 

dbooth: If they are removed, I'm assuming that we would tell people how to get the functionality other ways. 

Arthur: Who is using them? 
... If there are people who are using them, maybe we could get those samples. 

Kevin: In SAP we use them to describe client requirements. 

Scribe: ACTION: Kevin to write up example use of output operation and send to list. 

Jacek: Even though they are used, they seem to be used in different ways. 

TomJ: We didn't want to rewrite WSDL entirely, but we don't HAVE to redo everything in WSDL. Everybody seems to be using output operations in their own non-interoperable way. 
... So if we're going to burn too many cycles and can't come to consensus, then we could leave things as is. 

Arthur: This is also related to Requirement 85: Services need to be able to return a reference to another Service. 

JMarsh: But we don't have to have Service references to make output ops useful. 

Youenn: If we support MEPs in operations, we can specify these things in a somewhat more precise way. 

TomJ: Does anyone know what would be a good outcome of this? 

Arthur: I'd like to see some concrete examples of using output ops. 

<youenn> Youenn: We have 4 different types of operation but only 2 meps 

JeffS: I think that was the intention of the Soap over TCP example that I presented. Tom's point was that they're not CURRENTLY used that way. 

<Gudge> I thought we had 4 MEPs. Isn't each op type a type of MEP? 

<youenn> Youenn: So if we add mep support in operations, we should say what direction it is used and the trick is done 
... gudge: we have 4 types of operations implemented as two types of mep IMO 
... one way => input only & output only for instance 

Arthur: Right now there's a complete vacuum in the spec. 

JeffS: Some have strong concerns about how output op over HTTP could/should be done. 

TomJ: Do we want to pick one of the ways that people are currently doing it, or invent one of our own? It seems like people agree that they want them, but not how to standardize them. 

Gudge: By standardize, do you mean a standardized interpretation, such as "output only is an event notification"? 

TomJ: If WSDL is describing something, it is a contract and I'll know how to use it. 

Glen: You might standardize how I pass you a service reference so that you can do an output operation. 

Gudge: I think there's all sorts of ways people will use them. my preference is to allow them all. 

Glen: I think the problems need to be solved for the mainstream case. 

Prasad: I think the issue is not the binding, but the endpoint information. 
... That could be a part of contract negotiation. 
... It could be captured at the abstract level. 
... The only missing link is identifying the endpoint. 
... And we could provide examples of how they can be defined. 

JMarsh: So adding text saying that it can be out of band would be adequate? 
... And giving examples of how it can be given? 

Gudge: We can add a non-normative example. And we can add text describing that. 

JMarsh: Is there a concrete clarification proposal that would improve the understandability and interoperability of the output op? 

Gudge: Without a binding, nobody can use them. 

<GlenD> bindings make reality go 

Gudge: We could add prose that would describe how such a binding would work. 

JMarsh: If we take the TCP binding and clean it up and publish it as a note, we can refer to it from our spec to motivate the existance of the output op and explain how they can be used. 

JeffS: Could the people in favor of removing the output ops live with this approach? 

Jacek: That would satisfy me too. 

Roberto: I could live with a note also. 
... But several people have been using them in different ways. 
... So I'm okay with this approach as long as we provide some guidance. 

JMarsh: If we prohibit the unknown, would that help? 

Roberto: Yes. If we point people in the right direction, that would be okay. 

Arthur: I think it would be a big help, but if it's just TCP then it's not enough. I think we should have HTTP also. 

Scribe: ACTION: Arthur to propose an HTTP binding approach for output only 

Arthur: But it's related to R85, because it would be nice if we could say that the URL that's giving the notification declares another port type. 
... The only missing piece in WSDL is that there's no mechanism to say that a URL passed is the endpoint for a given binding. That's really the message of R85. 

Jacek: If we don't use output ops in HTTP binding, do we say explicitly that HTTP binding NEVER supports these ops? 

Arthur: No. No statement like that. 

Jacek: I think most of the confusion is that people use it in HTTP in different ways. 

Glen: Do you mean the HTTP GET binding or the HTTP Soap Binding? 

Jacek: I meant the WSDL Soap HTTP binding. 

Prasad: What's the issue with HTTP binding? The only issue I see is that the endpoint is given out of band. Everything else is the same. 

JMarsh: So people are more comfortable leaving output ops. If we also work on a note for TCP and look at the implications for HTTP binding (which Arthur is working on), what else do we need to do? Are we ready to resolve this? 
... I'm not entirely clear on the resolution. 
... In part 1, what needs to happen with part 1? 

Gudge: I think it's a part 2 thing. 

JMarsh: So in part 2, we would have a reference to this note. What would we say about it? 
... "Here's an example of how to do output ops"? 

Glen: What do we say in part 1? 
... Can someone specify any sequence of input/output ops? 

JMarsh: Let's consider that separately, as part of the MEP proposal. 

JeffS: I think for MEPs you're going to want the atoms of input and output. 

JMarsh: So in part 2 we say: "Here's an example of how you might use these" and refer to the note? 

Gudge: Yes. 

Arthur: I think we need an extension to have the strong type of URLs so that we know a given URL in a message refers to something in a binding. 
... If you give a type of "AnyURI", then you need to be able to reference this from a binding. 

Glen: A service reference? 

Arthur: Previously a service reference might have had more info about the service. This is just a URL. 
... A service reference is like a dynamically generated WSDL document. 

JMarsh: So we'll publish a note on this, reference the note from part 2, and you'll propose something for service references? 
... WOuld an out of band mechanism be allowed? 

Arthur: Yes, but it doesn't really describe the service. We need a standard way to do it. 

Scribe: ACTION: JeffS to see what would be needed to publish his TCP binding on output only as a Note. 

JeffS: I don't think it would be substantively different. 
... I thought TomJ and Jacek and Roberto were okay with closing this issue today. 

JMarsh: But Arthur volunteered to address the HTTP binding. 

Arthur: As far as part 1 is concerned, we just have to leave them in. 
... In part 2, we need clarification of what they output ops mean. 

JMarsh: Could we simply say that output ops are not used by some bindings? 

Arthur: Sure. 

Jacek: We also need to provide something that does use it. 

JMarsh: Yes, and we will still refer to a note of some kind. 

<GlenD> this issue seems to be to be a very slippery slope between low-level operations and application-level "orchestration" 

JMarsh: We could consider adding a TCP binding to the normative part, but I haven't heard anyone ask for it. 

<GlenD> I'm concerned that we're not considering all the ramifications, and haven't covered all the scenarios yet 

<youenn> agree 

<GlenD> (this is nothing new, 'twas ever thus, but this is one of the "known issues" with 1.1 which I think we're sort of punting on by closing this off now) 
... *shrug* 

JMarsh: I'm proposing that we state that HTTP does not support output-only, and then Arthur tries to remedy it. 

Scribe: ACTION: JMarsh to write up a proposed resolution for output-only ops and send it by email. 
... [Meeting adjourned] 


Summary of Action Items


ACTION: 2002-11-12: Glen and Paco to chase the Global Grid Forum WRT services implementing a single portType. -- -- PENDING 
ACTION: 2002-11-12: Paco will write two options for naming faults: schema vs WSDL. -- -- PENDING 
ACTION: 2002-11-12: Roberto will try and come up with another proposal for eliminating message, the discussion goes to email or the next f2f. -- -- PENDING 
ACTION: 2002-11-12: Marsh to contact the XMLP WG (and later the wider public) speaking about the issues around removing the use attribute. -- PENDING. 
ACTION: Arthur to propose an HTTP binding approach for output only 
ACTION: Glen to write up a description of the issues surrounding property description in WSDL. (This action replaces Glen's previous action item.) 
ACTION: JMarsh to write up a proposed resolution for output-only ops and send it by email. 
ACTION: JeffS to see what would be needed to publish his TCP binding on output only as a Note. 
ACTION: Kevin to write up example use of output operation and send to list. 
ACTION: dbooth to put URL for current minutes on group page [Done: http://www.w3.org/2002/ws/desc/#agenda ] 

  _____  

David Booth 
dbooth@w3.org
$Date: 2002/02/19 16:35:31 $